Cheating on Formula BOINC - The Scottish BOINC Team

[StefanR5R]

Cloud computing
Running multiple instances to get around 1000 task/day limits
Custom managers to get around 1000 task/day limits

None of these change credit/work.
Renting vs. owning is anybody's choice and primarily one of economics and availability, secondarily a matter of having learned how to operate remote hosts.
Having very many tasks in progress needs to be done wisely, to avoid degraded server performance.

Scripts that update update update, and steal all the GPU Grid and WCG GPU tasks

This doesn't change credit/work either; it only changes who receives work — or who succeeds in getting results reported — in situations with several different kinds of server-side bottlenecks. It wouldn't happen if server-side bottlenecks would not exist or would not matter much in contests... But sometimes they do. There is a grey area from gently forced updates to help a few high-throughput hosts getting on a level playing field with many low-throughput hosts on one end, to overdoing the update spam on the other end. These days, primitive update spam scripts are known to virtually all teams, and as far as I have read, made use of to questionable extent on occasion. Personally, I dislike contests for server attention and try to stay out of those.

Heck, I'm kinda impressed they found the loophole in the first place.

All it takes is a certain mindset. That there is a minimum quorum = 1 and that credit/result isn't fixed is plain to see, and there aren't many different variable credit schemes around. Now _who_ is thinking "Can I fool this to receive inflated credit?" when seeing this? And not just thinking it, but actually trying it?

Edit,

But I do agree with their punishment, all suspect points removed from the project. But perhaps they should be returned at calculated "fair value", as they DID complete those tasks.

True, the project admin has got a choice of how much he adjusts credit between 0 and normal. Likewise, the Formula BOINC admin has got a choice whether or not to react to what participants do, and whether or not to react if project admins changed credit after an FB contest finished...

 

[Icecold]

@StefanR5R said it better than I can, but I consider all the things listed a massive difference than inflating your work credit. All the things listed are just tactics to get more tasks and complete more work vs falsifying the work you actually did. How could anybody look at their stats page and not feel like a total scamming loser clown after knowingly falsifying the credit that they earned? I know I certainly couldn't.

 

[mmonnin03]

A client can only have 4 GPUGrid tasks no matter the number of updates a script can request.

They're only in it for the competitions. See their standings in the Marathon vs Sprints (5th vs 1st). And there has not been any kind of correction in the Sprint for the cheating so they basically got away with cheating. What happens when children are not corrected when doing something wrong, it keeps happening.

Edit: It looks like today they were DQ'd.

 

[Skillz]

A client can only have 4 GPUGrid tasks no matter the number of updates a script can request.

Not true. You can have more than 4 GPUGrid tasks on the stock client. My 5 and 6 GPU rigs can have more. Custom clients can have even more. It's just pointless to gobble up a bunch of those tasks at once since they have a huge bonus for returning them quick.

 

[mmonnin03]

4x per GPU per client. Happy now? My point still stands and is correct that you cannot download hundreds or thousands more by clicking update more often. Thats what I was referencing in Tonys post.